Peter Bailis: 2016 Plenary Session

10:45am

An increasing proportion of data today is generated by automated processes, sensors, and systems---collectively, the Internet of Things (IoT). A core challenge in IoT and an increasingly popular value proposition of many IoT applications is in identifying and highlighting unusual and surprising data (e.g., poor driving behavior, equipment failures, gunshots). We call this task---which is often statistical in nature and time-sensitive---analytic monitoring. To facilitate rapid development and scalable deployment of analytic monitoring queries, we have developed MacroBase, a new kind of data analytics engine that provides turn-key analytic monitoring of IoT data streams. MacroBase implements a customizable pipeline of outlier detection, summarization, and ranking operators. To facilitate efficient and accurate operation, MacroBase implements several cross-layer optimizations across robust estimation, pattern mining, and sketching procedures. As a result, MacroBase can analyze several million events per second on a single server. MacroBase has already uncovered several unexpected behaviors (and corresponding bugs) in production IoT deployments.

Peter will join Stanford University as an assistant professor of Computer Science. Peter received his Ph.D. in Computer Science from UC Berkeley in 2015 and an A.B. in Computer Science from Harvard College in 2011. His research in the Future Data Systems group (http://futuredata.stanford.edu/) focuses on the design and implementation of next-generation data-intensive systems.